Technology ❯ Hardware ❯ Chips
Google Cloud Nvidia
The push targets the CUDA lock-in that keeps PyTorch workloads on Nvidia GPUs.